How would I invert an unbalanced audio signal?

Recording Techniques, People Skills, Gear, Recording Spaces, Computers, and DIY

Moderators: drumsound, tomb

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

How would I invert an unbalanced audio signal?

Post by Professor » Mon Feb 13, 2006 2:10 pm

And before you go and state a bunch of obvious things here, let me put in a few requirements...
I'm thinking of a single-ended signal that has the A/C audio signal on the "positive" lead and a ground that is just simply a ground.
I know it's plenty easy to invert with a transformer, but I'd like a transformerless circuit.
I know that I can feed the signal into the 'wrong' side of an opamp as would be done in an electronically balanced output.

I'm curious if there is a way to invert the signal without a transformer or an opamp. Ideally, I'd like to know if there is a way to do it with passive electronics at all.

Application is actually kind of irrelevant because I'm looking at pure theory here while trying to come to a new and better understanding of phase & polarity. This quest may only be an exercise in semantics, but there's so much out there on the semantics of those two words that I would really like to put it to rest, at least in my mind.

So how about it? Can someone invert a single-ended signal without a transformer or opamp? or indeed without any active circuitry?

-Jeremy

User avatar
MD
gettin' sounds
Posts: 123
Joined: Wed Oct 15, 2003 11:37 am
Location: NY, NY

Post by MD » Mon Feb 13, 2006 3:02 pm

What if you just wire a cable backwards (tip to sleeve and vice versa)?

Am I overlooking something?

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

Post by Professor » Mon Feb 13, 2006 7:43 pm

Yes, you're overlooking that in an unbalanced cable the audio signal is carried as an alternating current wave down the center conductor while the sleeve should be a ground with no audio data - though it may have some electrical loading on it, but I'm not up to speed on the EE side of things which is why I'm asking.

-Jeremy

User avatar
MikeCzech
gettin' sounds
Posts: 127
Joined: Tue Nov 01, 2005 9:28 pm
Location: Salt Lake City, UT

Post by MikeCzech » Mon Feb 13, 2006 8:09 pm

Click on the little '0' in cubase.. Hehe, sorry..

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

Post by Professor » Tue Feb 14, 2006 1:56 pm

Yeah, I know it's easy enough with a plug in the digital world.
If I'm right about my guess here, then there may well not be a way to do it.
I mean an opamp is a reasonable solution, and it is used on the output of any balanced piece of pro gear. Indeed, two op amps are used and the signal is split and sent properly into one opamp and 'improperly' into the other so that it comes out inverted. Of course the active circuitry means a potential for added noise and distortion, but that's just chalked up to part of the overall sound of the device. That's part of the argument in favor of single-ended electronics given by audiophiles - less electronics in the signal path means greater fidelity, provided the single-ended runs are short enough.
On mixing consoles with "phase" switches, at least to my knowledge, those switches do not invert the single-ended signal. Rather they swap the positive and negative lines at the input, before the incoming balanced line is unbalanced. So in truth, they really are "phase reversal" switches because they swap the two phases (meaning two different conductors) of the incoming signal. Not unlike reversing the hot and neutral legs of a power system, these two 'phases' are reversed, but neither one has its 'polarity' inverted. No "polarity" inverting is involved because the circuit does not invert the polarity of a single-ended A/C wave - it swaps the 'correct' and 'incorrect' phases on the balanced line so they feed the wrong way into the opamp (or transformer) use to unbalance the line.

Unless someone knows another way this is done. If so, then please feel free to chime in.

-Jeremy

User avatar
Scodiddly
cryogenically thawing
Posts: 3616
Joined: Wed Dec 10, 2003 6:38 am
Location: Mundelein, IL, USA
Contact:

Post by Scodiddly » Tue Feb 14, 2006 4:31 pm

I don't think there is a way to invert polarity without a transformer or an active buffer. You're looking at electrons moving in a circuit - you need a different circuit where they can be moving in the opposite direction.

Well, there is a way - run everything balanced. Then it's easy to swap polarity.

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

Post by Professor » Tue Feb 14, 2006 5:19 pm

That's precisely my point Sco.
You're not inverting polarity, you are reversing the two phases on the balanced line as they enter the opamp that is unbalancing them. Internal to the equipment, everything runs unbalanced, even if the I/O is balanced.

-Jeremy

nclayton
steve albini likes it
Posts: 323
Joined: Sat Feb 11, 2006 8:08 am

Post by nclayton » Tue Feb 14, 2006 10:28 pm

I'm going to go ahead and say no, it's not possible to invert phase passively w/o a transformer of some kind. Well....a passive capacitor-inductor filter can theoretically do a 180 degree phase shift on an unbalanced signal, but only as long as there is no real resistance anywhere -- none in the source, the inductor's coils, the hookup wires, all of them have to have essentially 0 resistance, and to whatever extent there's resistance you'll get a frequency variable phase shift that would never quite be 180 degrees. With extreme component values you COULD actually get an essentially uniform very nearly 180 degree phase shift through the entire audio band, but you'd end up attenuating the signal like thousands or millions of times, and anyway the signal would be grossly frequency skewed even if you didn't mind the loss of level. So, I'm pretty sure that's not what you're looking for, and I'm PRETTY sure there's no other way.

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

Post by Professor » Tue Feb 14, 2006 10:55 pm

Actually that's precisely the kind of answer I was looking for. Not the grossly skewed signal, but that it can't be done.
I'm trying to work backwards through this confusion about polarity and phase.
You used the description 180? out of phase, which is a time-based measurement that is only accurate at one frequency. For example, a 180? phasse shift at 500Hz would be a delay of 1ms, not an inversion of the alternating current signal, and indeed at 1000Hz, that time-shift would be right back "in phase".
Usually this is where the 'polarity zealots' start stomping their feet and crying, "it's polarity, not phase!" And I used to be one of those, but I'm having second thoughts. I don't think we are actually inverting the polarity on our AC waves. Well at least, at the point when we think we are doing it, we are really reversing two different "phases" of the same AC signal, and then a device downstream is actually doing the inversion and our result may well be an inverted single-ended signal.
So, I'm kind of going back in time with this to try and figure out what our audio ancestors were thinking when they first labelled that switch a "phase reversal". I don't believe they had it wrong. Indeed, the more I'm understanding this, the more I think they are exactly right to call it a phase reversal that could only be done by taking advantage of some unique properties of either balanced audio lines or transformers.
I'm still holding out hope that someone may come along and correct me, and say, "no man, it's easy to invert a single-ended signal with an active circuit" or maybe even a glorious passive method. In that way, I'm an optimist. But I'm also cynical, because I'm starting to think they had it right way back then when they first drew in the little Φ symbol and called it phase reversal.

-Jeremy

User avatar
RodC
dead but not forgotten
Posts: 2039
Joined: Thu Dec 30, 2004 8:53 pm
Location: Right outside the door
Contact:

Post by RodC » Wed Feb 15, 2006 10:58 am

Ok, this idea looks OK on paper, and you would loose a LOT of gain, but consider this. Three resistors in series across your signal output. Connect your signal in across the resistor in the center - reversed polarity. Not sure how I could draw a diagram, or if you are even interested.


Input
+30
|
|
|
R1 3K
|
|
|----------Neg Out +15
|
R2 1.5K
|
|----------Pos Out +7.5
|
R3 1.5K
|
|
|
Negative


Theory, may be "total BS", and will change depending upon the impedance of the device connected...

If
R1 = 3K ohms
R2,3 are 1.5K ohms

Apply 30V to the input, the drop across R1 would be 15V. The V at the Pos Out would be + 7.5V, the V at the Neg out would be +15 (+7.5 V higher than the Pos, thus reversed)

Device indicated as input would see resistance of 1.5K and whatever the output device resistance is together parrallel. (The R1 3K and the R1&2 1.5K series are now in parrallel due to the ground of the Neg connection)

Change the R values to vary the voltage to taste.

Again may be "total BS", but Im sure someone else could chime in, but this will get us started. Maybe Brian will laugh at my answer and provide some insite LOL!

I thought at first a Wheatstone bridge could be used, you may want to look at some circuits using these for balanced inputs. Jenson transformers has one and the PAIA preamp uses one.

http://www.jensen-transformers.com/an/ingenaes.pdf

User avatar
cvanwinkle
gimme a little kick & snare
Posts: 98
Joined: Sun Jul 20, 2003 10:23 am
Location: Seattle, WA, USA
Contact:

Post by cvanwinkle » Thu Feb 16, 2006 9:10 am

Professor wrote:Actually that's precisely the kind of answer I was looking for. Not the grossly skewed signal, but that it can't be done.
I'm trying to work backwards through this confusion about polarity and phase.
You used the description 180? out of phase, which is a time-based measurement that is only accurate at one frequency. For example, a 180? phasse shift at 500Hz would be a delay of 1ms, not an inversion of the alternating current signal, and indeed at 1000Hz, that time-shift would be right back "in phase".
Usually this is where the 'polarity zealots' start stomping their feet and crying, "it's polarity, not phase!" And I used to be one of those, but I'm having second thoughts. I don't think we are actually inverting the polarity on our AC waves. Well at least, at the point when we think we are doing it, we are really reversing two different "phases" of the same AC signal, and then a device downstream is actually doing the inversion and our result may well be an inverted single-ended signal.
So, I'm kind of going back in time with this to try and figure out what our audio ancestors were thinking when they first labelled that switch a "phase reversal". I don't believe they had it wrong. Indeed, the more I'm understanding this, the more I think they are exactly right to call it a phase reversal that could only be done by taking advantage of some unique properties of either balanced audio lines or transformers.
I'm still holding out hope that someone may come along and correct me, and say, "no man, it's easy to invert a single-ended signal with an active circuit" or maybe even a glorious passive method. In that way, I'm an optimist. But I'm also cynical, because I'm starting to think they had it right way back then when they first drew in the little Φ symbol and called it phase reversal.

-Jeremy
I'm with you Jeremy. However, my Zealot side still pops up in the digital world. Great original question though. It had me reaching to the shelves and started looking at circuit diagrams again...

... aged like a fine cheddar

Professor
ghost haunting audio students
Posts: 3307
Joined: Wed May 07, 2003 2:11 pm
Location: I have arrived... but where the hell am I?

Post by Professor » Thu Feb 16, 2006 10:09 am

Oh yeah, in digital when you hit the 'invert' button it definitely inverts the waveform. But on the analog side of things, I'm definitely feeling that they had it right. Even as a bit of 'non-scientific' research, I took a look to see how a learned & experienced designer describes the function of that little Φ button, and sure enough, Rupert refers to it as a 'phase reversal'. Now I didn't try to contact him with the zealot argument, but I did find that interesting.

-Jeremy

nclayton
steve albini likes it
Posts: 323
Joined: Sat Feb 11, 2006 8:08 am

Post by nclayton » Thu Feb 16, 2006 9:41 pm

Professor wrote: You used the description 180? out of phase, which is a time-based measurement that is only accurate at one frequency. For example, a 180? phasse shift at 500Hz would be a delay of 1ms, not an inversion of the alternating current signal, and indeed at 1000Hz, that time-shift would be right back "in phase".
-Jeremy
Hi. You're right, if I was describing a circuit that produced a time delay it would only invert the phase at certain frequencies. But when I said 180 degrees out of phase I definitely wasn't using a time based description. If I was there would have been a "milliseconds" in there somewhere. I was talking about phase angle which is independent of frequency, so the theoretical circuit I was talking about would in fact invert polarity (=invert phase, =180 degree phase shift), not produce a specific constant time delay per se.

I don't know if this needs to be pointed out, but the resistive dropping circuit doesn't work because it doesn't produce a signal that is out of phase with the original signal compared to any useful reference. It's true that if we could think of the first dropped voltage as being the reference or "0", then the second dropped voltage would be out of phase with the original. When the signal is positive, the first dropped voltage is less positive and the third less positive than that, and the same for negative. So if you say |A|>|B|>|C| (which is true), and if you can define B=0 then obviously when A is positive C has to be negative and vice versa. But the problem is that if you actually were to ground that fist voltage node (B) then the second node (C) would be sitting between two grounded nodes and so it would have no signal on it anymore. Compared to the REAL ground, C is still in phase with A. If you can look at nodes B and C in a floating sense without any reference to a real ground then it works, but the only practical way to do that in real life is to use a transformer between those nodes and then you're back to where you started. This is just really long winded way of saying you might as well think of A as zero and let ground be "hot", which we know doesn't work.

Ned

nclayton
steve albini likes it
Posts: 323
Joined: Sat Feb 11, 2006 8:08 am

Post by nclayton » Thu Feb 16, 2006 9:52 pm

RodC wrote: I thought at first a Wheatstone bridge could be used, you may want to look at some circuits using these for balanced inputs. Jenson transformers has one and the PAIA preamp uses one.

http://www.jensen-transformers.com/an/ingenaes.pdf
I didn't read this whole article, but of course it's not a method for balancing an unbalanced signal, but rather a means to avoid accidentally unbalancing an already balanced or a floating signal.

nclayton
steve albini likes it
Posts: 323
Joined: Sat Feb 11, 2006 8:08 am

Post by nclayton » Thu Feb 16, 2006 11:16 pm

Professor wrote: I'm still holding out hope that someone may come along and correct me, and say, "no man, it's easy to invert a single-ended signal with an active circuit" or maybe even a glorious passive method. In that way, I'm an optimist. But I'm also cynical, because I'm starting to think they had it right way back then when they first drew in the little Φ symbol and called it phase reversal.

-Jeremy
Sorry I keep on just writing....I guess I just didn't quite finish reading your message before responding last time which I know is completely lame. But ANYWAY, it definitely IS super easy to invert a single ended signal with an active circuit using just one single transistor. The simplest possible active circuit, the common emitter amplifier, always inverts phase.

And by the way, the point of my first message was that theoretically you actually CAN invert phase passively, just not in a practically useful way.

As for the semantics of phase and polarity, I think there are some misunderstandings there, and probably not without reason. For one thing, phase can kind of be thought of as either a time shift or an angle shift, and it's not totally wrong to look at it from one standpoint or another. Most people when they talk about phase are talking about a change of angle of a signal with no time units attached, which is expressed mathematically as a complex number. On the other hand, in the more familiar "real world" equation of a wave like f(t)=C*sin(f*2*pi*t+Φ) where t is time, the part of the function that changes the phase (Φ) is really a time shift with units of seconds. But in a sense, in this equation Φ isn't really the "phase" of the signal, but rather, as you said, the phase is determined both by the variable Φ and the frequency f, so Φ controls phase, but isn't the phase angle itself.

I think it's fair to say phase inversion, 180 degree phase shift and polarity reversal all refer to the exact same thing and can be used as synonyms without reservation. They all just refer to flipping the signal upside down by making it negative. This is true even mathematically where multiplication by -1 is the same as a multiplication by i^2 which is the same as a 180 degree rotation if you're familiar with complex math. Of course with polarity you can only have two values, plus and minus, while the phase can be any number value under the sun, so basically polarity reversal is just a specific subset of general phase change.

It's true that in many (but definitely not all) consoles the phase reverse switch just reverses the leads to a balanced input circuit like you could do if you switched the phases of a two phase AC power supply. But what's important isn't so much how it's done, but what's actually done. Consoles that use a CE amplifier or an opamp to reverse the signal polarity (or digital audio programs that multiply the value by -1) can still call what they're doing phase reversal without abusing language. After all, we all know what a phase inverter in a guitar amplifier does, but it doesn't work by swapping leads in a balanced circuit, or by producing a time delay, but by the natural way that the common emitter (or usually in this case the common cathode) amplifier inverts polarity.

Whatever....

Ned

Post Reply

Who is online

Users browsing this forum: No registered users and 53 guests